Background:  We randomized 130 ART-naive asymptomatic patients, with 300 to 500 CD4/μL, to receive either interleukin (IL) -2 (n = 66) or no treatment (n = 64). IL-2 was given at weeks 0, 8, 16, and 24; 1 to 2 additional cycles were administered to 34 and 4 patients during year 2 and 3, respectively. Over 96 weeks, 36% patients in the IL-2 arm and 61% in the control arm (p = 0.005) progressed to the study end-point, defined as a confirmed CD4 count of 300 cells/μL or less, the initiation of ART, or an AIDS-defining event or death. In the IL-2 arm, 6 AIDS-related events or death occurred and 5 in the control arm. Follow-up was extended over 150 weeks.

Methods:  To identify predictors of progression to study end-point, age, gender, treatment arm, baseline CD4 count, and plasma viral load were included in Cox models. Medians were used as cut-offs in Kaplan Meier stratified analyses, except for plasma viral load which used a 4.5 log₁₀ copies/mL cut-off.

Results:  At baseline, median age was 36 years, CD4 383 cells/μL, and plasma viral load 4.36 log₁₀ copies/mL. Predictors were analysed separately in each arm because of a strong interaction (p = 0.006) between plasma viral load and treatment arm. In controls, baseline CD4 was the only variable significantly (p <0.0001) associated with progression to study end-point. In the IL-2 arm, progression was associated with plasma viral load (HR = 3.7, p = 0.0006) and CD4 (HR = 0.99, p = 0.01). Through week 150, 34 of 65 (52%) patients in the IL-2 arm and 52 of 63 (83%) in the control arm progressed to the study end-point (p <0.001) without any new AIDS-defining event or death beyond week 96. In the plasma viral load <4.5 log₁₀ stratum, estimates (95%CI) of the probability of non progression through week 150 were 0.65 (0.48 to 0.78) in the IL-2 arm and 0.10 (0.02 to 0.26) in the control arm (p <0.0001). Corresponding figures in the >4.5 log₁₀ stratum were 0.06 (0.01 to 0.24) and 0.11 (0.02 to 0.27) (NS). Similarly in the plasma viral load <4.5 log₁₀ stratum, the probability of not being on ART through week 150 were 0.72 (0.55 to 0.84) and 0.16 (0.04 to 0.34) in the IL-2 and control arms respectively (p <0.0001), with an observed delay of 92 weeks of 1^st quartile time to ART initiation.

Conclusions:  Our data show that IL-2 alone could prevent CD4 count decline for years in ART-naive patients with a plasma viral load <4.5 log₁₀ copies/mL. Long-term efficacy and clinical safety of such a strategy should be further investigated.